This paper has been withdrawn by Md Anwar Hossen
No PDF available, click to view other formats
Abstract:Continual learning in Large Language Models (LLMs) is hindered by the plasticity-stability dilemma, where acquiring new capabilities often leads to catastrophic forgetting of previous knowledge. Existing methods typically treat parameters uniformly, failing to distinguish between specific task knowledge and shared capabilities. We introduce Mixture of Sparse Experts for Task-Agnostic Continual Learning, referred to as SETA, a framework that resolves the plasticity-stability conflict by decomposing the model into modular subspaces. Unlike standard updates, where tasks compete for the same parameters, SETA separates knowledge into unique experts, designed to isolate task-specific patterns, and shared experts, responsible for capturing common features. This structure is maintained through elastic weight anchoring, which protects critical shared knowledge and enables a unified gating network to automatically retrieve the correct expert combination for each task during inference. Extensive experiments across diverse domain-specific and general benchmarks demonstrate that SETA consistently outperforms state-of-the-art parameter-efficient fine-tuning-based continual learning methods.
| Comments: | we are updating the paper and will release another version soon |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2601.17616 [cs.LG] |
| (or arXiv:2601.17616v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2601.17616 arXiv-issued DOI via DataCite |
From: Md Anwar Hossen [view email]
[v1]
Sat, 24 Jan 2026 22:39:22 UTC (941 KB)
[v2]
Fri, 1 May 2026 18:20:49 UTC (1 KB) (withdrawn)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。